自定义数据集问题

[yfq512]

Prerequisite

• I have searched Issues and Discussions but cannot get the expected help.
• I have read the FAQ documentation but cannot get the expected help.
• The bug has not been fixed in the latest version (dev-1.x) or latest version (dev-1.0).

Task

I'm using the official example scripts/configs for the officially supported tasks/models/datasets.

Branch

main branch https://github.com/open-mmlab/mmdetection3d

Environment

all is ok

Reproduces the problem - code sample

train

Reproduces the problem - command or script

我参照s3dis数据集制作了自己的数据集，就是根据一个大场景的点云（xyzrgb），识别里面的电线杆和路灯。一开始我只标注了电线杆和路灯，训练和测试发现，背景也会被误识别成电线杆或者路灯。之后我增加了背景类（事实上我只想识别出电线杆和路灯，为了让背景不乱识别，所以单独给他一个分类），当再次训练时，我发现背景类的准确率很高，而电线杆和路灯准确率为0，
+---------+-------------+--------+--------+--------+--------+---------+
| classes | dianxiangan | ludeng | bg | miou | acc | acc_cls |
+---------+-------------+--------+--------+--------+--------+---------+
| results | 0.0000 | 0.0000 | 0.9899 | 0.3300 | 0.9899 | 0.3333 |
+---------+-------------+--------+--------+--------+--------+---------+

我猜可能是背景类的点云数量占比太大造成的，怎么解决这类问题呢？
此外，我的数据是自己标的因此数据量较少，不确定这种情况和数据量是否也有关系

Reproduces the problem - error message

1

Additional information

1